A knowledge graph–based network framework for systemic risk assessment in the low-altitude economy industrial chain

This study develops a knowledge graph–based framework for systemic risk assessment in the low-altitude economy industrial chain. An LLM integrates heterogeneous industrial information into a graph comprising 92 entities and 134 relations. Weighted degree centrality is combined with CoVaR and MES to construct KN-ΔCoVaR and KN-MES. Simulations, sensitivity analyses, and case studies show that these indicators outperform traditional measures in identifying critical enterprises and assessing systemic losses. Highly connected enterprises can amplify risk despite low standalone financial risk. The framework supports interpretable risk monitoring and early warning.